A Comparison of the Performance of Fama-French Multifactor Asset Pricing Models: An Application on Borsa İstanbul
Güler Aras, İlhan Çam, Bilal Zavalsız, Serkan KeskinExploring the factors that explain changes in stock returns is one of the most important research topics in finance literature. Recently, Fama and French have developed the five-factor asset-pricing model (FF5F) by adding profitability and investment factors on the three-factor model (FF3F), which consists of market, size and value factors. This model has been tested in various developed countries, especially in the USA, and has proved its success in explaining the changes in stock returns. However, there are deficiencies in the researches on whether this model is valid for developing countries with different dynamics from developed countries. In this paper, it is aimed to examine whether the FF5F is valid for Turkish stock market and to test how successful the FF5F is in comparison with the CAPM, FF3F and other alternative models. For this purpose, returns of 18 different intersection portfolios have been analyzed during the period of 150-months between January 2005 and June 2017. According to the regression results of mean absolute values of intercept terms, mean adjusted R-squared values, GRS-F test statistics and its p-values, it has been found that FF5F performs better than the other alternative models in the Turkish stock market.
Fama-French Çok Faktör Varlık Fiyatlama Modellerinin Performanslarının Karşılaştırılması: Borsa İstanbul Üzerine Bir Uygulama
Güler Aras, İlhan Çam, Bilal Zavalsız, Serkan KeskinHisse senedi getirilerindeki değişimi açıklayan faktörlerin neler olduğunun ortaya koyulması, finans literatüründeki önemli araştırma konuları arasında yer almaktadır. Bu bağlamda, Fama ve French, piyasa, büyüklük ve değer faktörlerinden oluşan üç faktörlü varlık fiyatlama modeline (FF3F), kârlılık ve yatırım faktörlerini de ekleyerek beş faktörlü bir varlık fiyatlama modeli (FF5F) geliştirmişlerdir. Bu model, ABD başta olmak üzere, çeşitli gelişmiş ülke piyasalarında test edilmiş ve modelin getirilerdeki değişimin açıklanmasındaki başarısı kanıtlanmıştır. Ancak bu modelin, gelişmiş ülke piyasalarından farklı dinamiklere sahip olan gelişmekte olan ülke piyasalarında geçerli olup olmadığı ile ilgili araştırmalarda eksiklikler vardır. Bu çalışmada, Türkiye hisse senedi piyasası için, FF5F’nin geçerli olup olmadığının incelenmesi ve FF5F’nin CAPM ve FF3F başta olmak üzere diğer alternatif modellere göre ne kadar başarılı performans gösterdiğinin test edilmesi amaçlanmıştır. Bu amaç doğrultusunda, Ocak 2005 - Haziran 2017 tarihleri arası 150 aylık dönemde, 18 adet kesişim portföyünün getirileri üzerinden analizler yapılmıştır. Regresyon analizlerinden elde edilen sabit terimlerin mutlak değerlerinin ortalaması, ortalama düzeltilmiş R2 değerleri, GRS–F test istatistik ve p-değeri sonuçları değerlendirildiğinde, FF5F’in Türkiye hisse senedi piyasasında diğer alternatif modellerden daha iyi performans gösterdiği bulgusuna ulaşılmıştır.
Exploring the factors that explain changes in stock returns is one of the most important research topics in finance literature, and thus scholars have introduced a wide variety of models. The most well-known is the capital assets pricing model (CAPM), which explains the change in returns with a single factor, called the market beta measuring the systematic risk of the market. The most important feature expected from any asset-pricing model is the power to explain the change in stock returns. Many researchers have found that the power of CAPM is not sufficient. Thus, alternative asset pricing models have been proposed to develop CAPM. Among these models, the three-factor asset pricing model offered by Fama and French has been tested by many researchers in both developed and developing countries. It has been proven that the performance of the model is better than CAPM. Fama and French’s threefactor model explain changes in stock returns with market, size and value factors. After three-factor models, Fama and French has been criticized about the lack of their models. With the discovery of several factors that has any information in explaining returns (such as momentum, volatility, liquidity, investment and profitability factors), the use of multi-factor models has become widespread in explaining stock returns. In line with these advances in the literature, Fama and French acknowledged that a significant portion of the change in stock returns could not be explained by the three-factor model. Thus, Fama French have introduced five-factor model by adding profitability and investment factors to three-factor model. They were based on the dividend discount model when making decisions to add only these two factors to the three factor models. The Five-factor model has been tested in various developed countries, especially in the USA, and has proved its success in explaining the changes in stock returns. However, there are deficiencies in the researches on whether this model is valid for developing countries with different dynamics from developed countries (such as higher returns and higher price fluctuations). In this paper, it is aimed to examine whether the five-factor model is valid for the Turkish stock market and to test how successful the five-factor model is in comparison with the CAPM, three-factor model and other alternative three and four factor models. For this purpose, average monthly percent excess returns for eighteen intersection portfolios formed on six Size and Value, six Size and Profitability, six Size and Investment have been analyzed during the period of 150-months between January 2005 and June 2017. Regression results of mean absolute values of intercept terms, mean adjusted R-squared values, GRS-F test statistics and its p-values were used to evaluate the relative performance of the models in capturing the changes in the returns of the left-hand-side intersection portfolios in the Turkish stock market. It has been found that FF5F performs better than the CAPM, the three-factor model and other alternative three - or four-factor models in the Turkish stock market. The empirical findings in the paper conform to the findings obtained in both developed and developing countries. Moreover, these findings are also similar to those obtained from other studies testing the five-factor model on BIST and using a different sample period and different portfolio sets from this paper. Future studies can be focused on to develop a more comprehensive asset pricing model by adding other factors that have been proven to explain the average returns (such as momentum factor, inverse momentum factor, liquidity factor, volatility factor) to the five-factor model. Since what is most important here is to find the model that explains more change with less factors, it will be substantial to identify unnecessary factors and to have an asset-pricing model that shows more effective results for developing countries such as Turkey.
